Blue Cross Ties $942 Million in Added Costs to Hospitals’ AI Coding, while Hospitals Question the Analysis’s Accuracy

In September 2026, the Blue Cross Blue Shield Association (BCBSA) published an analysis estimating that hospitals’ use of AI-enabled coding tools added $942 million to Blue System costs in 2024 and 2025 compared with 2023 levels.

On October 5, the American Hospital Association (AHA) responded that the report draws conclusions its data can’t support.

Payers argue the coding changed, but the care didn’t.

Reviewing claims from early 2023 to late 2025, BCBSA identified more than 55,000 additional complex cases where an added secondary diagnosis bumped the claim into a higher-paying category. Many of the secondary diagnoses, such as anemia, can be identified from a single lab result, the kind of detail AI tools are built to catch.

The report’s core evidence is a gap between diagnosis and treatment. For example, hospitals with the fastest growth in complex coding diagnosed anemia more often than other hospitals but gave blood transfusions (a common anemia treatment) to a smaller share of those patients. BCBSA reads that gap as evidence that coding outpaced care.

“If patients are truly sicker, we’d expect to see more treatment,” said Luke Chalker, a BCBSA senior vice president.

Hospitals argue the analysis can’t distinguish accurate coding from inflated coding.

The AHA contends that more diagnoses aren’t evidence of bad coding. Americans are living longer and managing more chronic conditions, while lower-acuity care continues to exit the hospital. To the AHA, a rise in complex inpatient claims is expected.

Hospitals also dispute using treatment as the yardstick. A condition like anemia may not call for a transfusion but can still shape clinical decisions, nursing needs, or medication management. Because those effects don’t always show up as separate services, the AHA argues the right question isn’t whether reimbursement increased, but “whether the patient’s medical record supports the diagnosis.”

The AHA adds that BCBSA’s methods can’t answer that question. The report didn’t review medical records to determine whether diagnoses were clinically supported, and it didn’t appear to identify which claims involved AI tools rather than traditional coding.

Health system leaders expect revenue cycle AI investment to increase and pay off.

In THMA’s 2026 Technology Investment Survey, 39% of health system technology leaders expect revenue cycle and financial AI technologies to see the largest increase in investment over the next 6–12 months, the highest of any category.

Similarly, our 2026 Finance and Budgeting Survey shows 84% of health system CFOs and VPFs expect revenue cycle and financial technologies to generate the most significant AI ROI over the next 12 months.

The BCBSA report could raise the bar for what counts as a return: revenue captured through AI coding only holds if it stands up to payer review.

So What?

• Hospitals face growing pressure to show that the medical record supports what they bill.

• More payer scrutiny is likely. This is BCBSA’s second such analysis, and BCBSA has called for clear expectations for hospitals using AI coding tools.

• Health systems face a dilemma: revenue cycle is where many leaders expect AI’s biggest return, but aggressive use could invite payer pushback. Many systems are still determining how much AI-assisted coding they can defend to payers.